clear
version 8.0
#delimit ;
cd "Z:\interactionmodels\";                       * SET WORKING DIRECTORY;
log using results\unconditionalfragmentation.log, replace;

*     ***************************************************************** *;
*     ***************************************************************** *;
*       File-Name:      unconditionalfragmentation.do                   *;
*       Date:           01/09/2007                                      *;
*       Author:         MRG                                             *;
*       Purpose:        Check MSG claims about the unconditional effect *;
*                       of Fragmentation                                *;
*       Input File:     STATA_mozaffar.dta                              *;
*       Output File:    unconditionalfragmentation.log                  *;
*       Data Output:    none                                            *;
*       Previous file:                                                  *;
*       Machine:                                                        *;
*     ****************************************************************  *;
*     ****************************************************************  *;

set mem 10m;

use getdata\STATA_mozaffar.dta;


*     ****************************************************************  *;
*     ****************************************************************  *;
*           Mozaffar et al. report some things in footnotes that we     *;
*           should check.                                               *;
*     ****************************************************************  *;
*     ****************************************************************  *;


*     ****************************************************************  *;
*       First, fragmentation independently reduces the number of        *;
*       electoral and legislative parties.  Footnote 7 says that this   *;
*       comes from Model 2 and Model 3 but without the interaction term.*;
*       Let's take a look.                                              *;
*     ****************************************************************  *;

regress  elecparties fragmentation2 concentration, robust;

regress  elecparties fragmentation concentration, robust;

*     ****************************************************************  *;
*       Fragmentation has a negative but not significant effect on      *;
*       electoral parties.                                              *;
*     ****************************************************************  *;

regress  legparties fragmentation2 concentration, robust;

regress  legparties fragmentation concentration, robust;

*     ****************************************************************  *;
*       Fragmentation has a positive but not significant effect on      *;
*       electoral parties.                                              *;
*     ****************************************************************  *;

regress  elecparties fragmentation2 concentration logmag10 proximity prox_prescandidate, robust;

regress  elecparties fragmentation concentration logmag10 proximity prox_prescandidate, robust;

regress  legparties fragmentation2 concentration logmag10 proximity prox_prescandidate, robust;

regress  legparties fragmentation concentration logmag10 proximity prox_prescandidate, robust;

*     ****************************************************************  *;
*       Thus, there is no evidence to support the claim that            *;
*       fragmentation independently reduces the number of electoral and *;
*       legislative parties.                                            *;
*     ****************************************************************  *;

log close;
exit;
